Add UTF8Encoder contract (#5)

devstein · web-flow · commit d981965aca9f · 2022-01-16T13:54:29.000-08:00
diff --git a/contracts/UTF8Encoder.sol b/contracts/UTF8Encoder.sol
@@ -0,0 +1,42 @@
+//SPDX-License-Identifier: MIT
+pragma solidity ^0.8.9;
+
+/// @title A library for encoding UTF-8 strings
+/// @author Devin Stein
+library UTF8Encoder {
+  /// @notice Get the UTF-8 string for `self`
+  /// @dev encode will error if the code point is not valid
+  /// @param self The code point to encode
+  /// @return The UTF-8 string for the given code point
+  function encode(uint32 self) external pure returns (string memory) {
+    bytes memory out;
+    if (self <= 0x7F) {
+      // Plain ASCII
+      out = bytes.concat(bytes1(uint8(self)));
+      return string(out);
+    } else if (self <= 0x07FF) {
+      // 2-byte unicode
+      bytes1 b0 = bytes1(((uint8(self) >> 6) & (uint8(0x1F))) | (uint8(0xC0)));
+      bytes1 b1 = bytes1(((uint8(self) >> 0) & (uint8(0x3F))) | (uint8(0x80)));
+      out = bytes.concat(b0, b1);
+      return string(out);
+    } else if (self <= 0xFFFF) {
+      // 3-byte unicode
+      bytes1 b0 = bytes1(uint8(((self >> 12) & 0x0F) | 0xE0));
+      bytes1 b1 = bytes1(uint8(((self >> 6) & 0x3F) | 0x80));
+      bytes1 b2 = bytes1(uint8(((self >> 0) & 0x3F) | 0x80));
+      out = bytes.concat(b0, b1, b2);
+      return string(out);
+    } else if (self <= 0x10FFFF) {
+      // 4-byte unicode
+      bytes1 b0 = bytes1(uint8(((self >> 18) & 0x07) | 0xF0));
+      bytes1 b1 = bytes1(uint8((self >> 12) & 0x3F) | 0x80);
+      bytes1 b2 = bytes1(uint8(((self >> 6) & 0x3F) | 0x80));
+      bytes1 b3 = bytes1(uint8(((self >> 0) & 0x3F) | 0x80));
+      out = bytes.concat(b0, b1, b2, b3);
+      return string(out);
+    }
+    require(false, "invalid unicode code point");
+    return "";
+  }
+}
diff --git a/scripts/contracts.ts b/scripts/contracts.ts
@@ -2,31 +2,21 @@ import { ethers } from "hardhat";
 import { Contract } from "ethers";
 
 const UNICODE_LIBRARY = "Unicode";
+const UTF8_ENCODER_LIBRARY = "UTF8Encoder";
 const UNICODE_DATA_CONTRACT = "UnicodeData";
 
-export const deployUnicodeData = async (): Promise<Contract> => {
-  const contractFactory = await ethers.getContractFactory(
-    UNICODE_DATA_CONTRACT
-  );
+export const deploy = async (id: string): Promise<Contract> => {
+  const contractFactory = await ethers.getContractFactory(id);
   const contract = await contractFactory.deploy();
   await contract.deployed();
-  console.log(
-    `${UNICODE_DATA_CONTRACT} contract deployed: ${contract.address}`
-  );
-  return contract;
-};
-
-export const deployUnicodeLibrary = async (): Promise<Contract> => {
-  const contractFactory = await ethers.getContractFactory(UNICODE_LIBRARY);
-  const contract = await contractFactory.deploy();
-  await contract.deployed();
-  console.log(`${UNICODE_LIBRARY} contract deployed: ${contract.address}`);
+  console.log(`${id} contract deployed: ${contract.address}`);
   return contract;
 };
 
 export const deployAll = async (): Promise<Contract[]> => {
-  const unicodeData = await deployUnicodeData();
-  const unicodeLibrary = await deployUnicodeLibrary();
+  const unicodeData = await deploy(UNICODE_DATA_CONTRACT);
+  const unicodeLibrary = await deploy(UNICODE_LIBRARY);
+  const utf8Encoder = await deploy(UTF8_ENCODER_LIBRARY);
 
-  return [unicodeData, unicodeLibrary];
+  return [unicodeData, unicodeLibrary, utf8Encoder];
 };
